Query Chain Analysis for Search Relevance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches to analyzing user data for content delivery are inefficient and inaccurate, often resulting in biased models that fail to optimally reflect user interests, leading to ineffective targeting of relevant content or advertisements.
Innovation Solution
The implementation of query chain analysis to identify patterns in user search queries and transactions, which involves detecting and annotating query chains to improve product search relevance and predict user behavior, by propagating positive feedback from previous related queries and incorporating query chain information into behavioral models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If large amounts of user data are stored and analyzed using a single computer, then processing power is sufficient, but the approach is costly and inefficient
Solution Approach 1:
The patent segments user data into distinct blocks or categories (e.g., by user behavior patterns, query types, or time periods) and processes these segmented data sets in parallel using multiple computing resources. This allows the system to maintain high processing accuracy through comprehensive analysis while improving efficiency through distributed parallel processing, directly resolving the contradiction between reliability and productivity.
2Productivity
If individual models are averaged or merged, then data is processed in digestible blocks, but unnecessary variances and biases are introduced
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors the performance and predictions of individual behavioral models, identifies sources of variance and bias, and uses this feedback to refine and adjust the models before aggregation. This ensures that when models are combined, the variances and biases are minimized through iterative optimization, maintaining high measurement precision while enabling parallel processing of multiple models.
3Ease of operation
If conventional approaches associate users with categories based on single interactions, then implementation is simple, but user interests are not optimally reflected
Solution Approach 1:
The patent applies preliminary action by analyzing and weighting multiple user interactions before final category assignment. Instead of简单地 associating users with categories based on single interactions, the system pre-processes historical data, assigns weights to different interaction types based on their predictive value, and aggregates these weighted interactions to determine user categories. This preliminary weighting and aggregation process improves user interest accuracy while maintaining implementation feasibility through systematic data processing.
Data Source
AI summary
Embodiments of the present invention provide improved techniques for determining long term relevance and user behavior using query chains. The query chains may first be detected and then annotated into different types of chains based at least in part on various decision rules, machine-learned classifiers, and inter-query relationships. The query chains may then be subsequently used to train models for predicting user behavior and providing more relevant results to a user's queries. A content provider system according to various embodiments may aggregate historical data associated with previous search and/or transaction data, which may be analyzed to detect query chains, for example, whether queries are chained to each other. Determining whether queries are chained to each other may involve incorporating decision rules and reformulation models, analyzing temporal windows between queries, and/or analyzing inter-query relationships.